Evidence from several sources points toward the existence of latency by JC Virus (JCV) in the CNS in a significant portion of the human population. This includes: (a) our own studies of progressive multifocal leukoencephalopathy (PML) pathology. (b) The appearance of PML in 2-5% of autopsied patients with acquired immunodeficiency syndrome (AIDS). (c) Our evidence from the hamster model of JCV CNS infection, in which JCV inoculated intracerebrally in the neonatal hamster brain localizes in the cerebellar granular layers, hippocampus and periventricular areas. This evidence suggests a mechanism by which low levels of JCV entering the brain through the circulation early in life might become focally distributed through replication as a minichromosome in synchrony with host cell division. (d) Preliminary evidence from studies of normal brain and brain with neurological disease other than multiple sclerosis (MS) using the polymerase chain reaction (PCR) for amplification of JCV DNA suggests that the virus might be present in a latent state in brains without overt demyelinating disease. The weight of evidence in favor of latency by JCV in the brain has led us to formulate a theory of pathogenetic mechanisms by which latent and partially reactivating (limited to early region expression or other restricted expression without independent DNA replication) might lead to immune-mediated demyelinating disease, such as seen in MS. It is proposed that PML and MS may be the result of differing host responses to the presence of virus focally distributed in the brain following infection of glial cells or their precursors early in life. While JCV is a candidate for involvement in both diseases, other DNA viruses with some specificity for infection of and expression in glial cells should also be considered.